Capital Requirements for Entry into Property and Liability Underwriting: An Empirical Examination

Capital Requirements for Entry into Property
and Liability Underwriting: An Empirical
Examination
J. D. Hammond and Arnold F. Shapiro*
Regulation of insurers has focused primarily on insurer solidity.1 The
concern for solidity centers on safe capitalization requirements for insurers,
premium rate adequacy, and investment controls. Although current regulatory
concerns frequently address issues of pricing equity and new approaches to risk
classification, solidity remains the principal consideration.
Capital adequacy, a key solidity requisite, has generally been addressed from
a context of rules-of-thumb and general conservatism. Questions of capital
adequacy are relatively complex and analysis of risk theory and finance has been
difficult to translate into statutes and difficult for insurance commissiohers to
incorporate into regulatory practice. Rule.of-thumb analysis, therefore, has held
considerable appeal for insurance commissioners.
In recent years, research into the capital adequacy of established insurers
has expanded considerably. Aside from the long-standing conceptual contributions of risk theory, advances were stimulated greatly by the linking of portfolio
theory to the risk and return analysis of different lines of insurance.2 The most
thorough explanation is the work of Bachman, who employed the optimization
features of the portfolio model to estimate minimum capital requirements
1 Most dimensions of insurance regulation are concerned with insurer solvency. However, solidity is a broader, although less precise, term than solvency and implies a higher
standard of surveillance and concern than just solvency alone. The term was first used in
the European literature and its acceptance into the U.S. literature on insurer regulation has
resulted largely from the writings of Spencer L. Kimball. See, Spencer L. Kimball, "The
Purpose of Insurance Regulation: A Preliminary Inquiry in the Theory of Insurance Law."
Minnesota Law Review, March 1961, especially pp. 478-486.
~See J. Robert Ferrari, "A Theoretical Portfolio Selection Approach for Insuring
Property and Liability Lines," Proceedings of the Casualty Actuarial Society, 1967, pp.
33-54. So far as we can tell, Ferrari was the first to apply the Markowitz portfolio analysis
in an insurance underwriting context.
*J.D. Hammond is Professor of Business Administration and Arnold F. Shapiro is
Associate Professor of Business Administration, both at Pennsylvania State University. This
paper is based upon a larger study supported by a grant from the National Science Foundation and reported in detail in the Technical Report, The Regulation of Insurer Solidity
through Capital and Surplus, APR 75-16550 A01. Several students assisted in the original
project and in the preparation of this paper. They were: Deborah Cart, William Daniels,
Mark Gruskin, Bruce Leidenberger, John Markley, John McAdon, and Joseph Stunja. The
authors are grateful to them. Any opinions, findings, conclusions, or recommendations
arising from that study are those of the authors and do not necessarily reflect the views of
the National Science Foundation.
225
226
REGULATION OF FINANCIAL INSTITUTIONS
associated with optimal combinations of insurance lines, grouped in such a way
as to minimize the variance or risk of the entire underwriting portfolio.3 The
National Association of Insurance Commissioners has also been active in
developing and refining its "Early Warning System," a series of financial tests
designed to identify insurers with questionable solidity. Our own work has also
explored the application of portfolio theory towards the estimation of capital
required for conduct of the underwriting function.4 Empirical research dealing
with the capital required for entry into the various lines of insurance, however,
has been sparse.
The purpose of the present paper is to examine empirically the entry capital
for nonlife insurers required to conduct the insurance underwriting function, to
estimate differences in underwriting risk among insurance lines which can be
associated with entry level underwriting activity, and to develop information
which enhances the ability of regulators to assess entry risk.s The paper first
discusses the significance of the entry topic and summarizes current capital
requirement statutes. Then, industry-wide data are employed to prepare estimates of the underwriting risk at entry for the interval of the study, 1972-1975.
Lines which were consistently of high risk are identified and estimates are made
of the maximum ratios of premium volume to capital which can safely be
accommodated in each line of insurance. Finally, suggestions are made for
improvement of entry capital requirement statutes.
Entry Capital Requirements6
Most of the statutes specifying entry capital requirements are straightforward, simply indicating an absolute amount to be met as a condition for
licensure. Most were almost certainly formed judgmentally without reference to
~ James E. Bachman, Capitalization Requirements for Multiple Line Property-Liability
Companies, Huebner Foundation Monograph No. 6, (Homewood, Illinois: Richard D. Irwin
Co., 1978). The bibliography of the monograph provides a thorough listing of publications
dealing with insurer financial analysis and also with its links to portfolio theory.
4J.D. Hammond and Ned Shilling, "Some Relationships of Portfolio Theory to the
Regulation of Insurer Solidity," Journal of Risk and Insurance, September 1978; also
reprinted by the American Bar Foundation as Research Contributions of the American
Bar Foundation, 1978, No. 2. Also, a part of the National Science Foundation Study on
which the current paper is based examined underwriting capital needs in an empirical,
portfolio theory context. Technical Report NSF Grant Apt75-16550 A01, The Regulation
of lnsurer Solidity through Capital and Surplus Requirements. One of the difficulties in
application of the theory to underwriting portfolios is that variances of lines of insurance
do not remain constant as premium volume changes.
s Established insurers generally face a different and probably more stable set of risks
than new insurers. The beneficial pooling effects of large numbers are more likely to be
present, the expense drain of start-up costs has disappeared, market niches and identities
are apt to be present, with survival implying at least some ability to withstand competitive
and perhaps regulatory constraints.
6 Unless otherwise noted, the word capital is used ttL~oughout the study to denote both
capital and surplus. We are grateful to the National Association of Independent Insurers for
supplying a summary of the capital requirement laws of all states. The compilation is
complete into 1976.
PROPERTY UNDERWRITING CAPITAL
HA MMO ND- SHA PIR 0
227
any formal research techniques. The capital requirements of New York State
were enacted in 1939 and have not been appreciably altered since.7 A recently
proposed revision would simply double the requirements of ’the existing New
York law.
Entry capital requirements also reflect some balance between concern over
insurer solidity and the marketplace benefits associated with ease of entry. The
emphasis is not uniform across states, however. The Insurance Department of
Illinois, for example, described the problem:
The minimum capital and surplus requirements were substantially increased
to their present levels in 1971. It is doubtful legislation to require additional
capital and surplus... One of our major concerns and responsibilities is to
encourage competition among insurers. Such competition is largely dependent upon ease of entry into the Illinois market. Excessive or unreasonable
statutory capital and surplus requirements would, in the long run, stifle that
competition by discouraging the growth of new companies in Illinois.8
Illinois requirements for underwriting on a multiple-line basis amount to
$1,000,000 of capital and surplus. On the other hand, those of New York
specify approximately $6,000,000. Nonetheless, the New York Department
states that:
It is now practic’ally impossible for a new company financed on the
minimum basis (required by law) to maintain adequate financial strength
and to write sufficient volume to absorb the necessary overhead expenses
and leave an underwriting margin after losses. The low amounts presently
required by law have permitted the formation of companies which have
encountered considerable difficulty in maintaining proper financial strength
to meet their commitments.9
No more is required to illustrate the varying philosophies among states concerning entry capital requirements.
Market Importance of Entry
In the last 10 years, 390 new property and liability insurers have been
formed. The details are shown in Table 1.
Of the 390 new formations, 321 were stock insurers which attracted a total
of $709,367,000 from investors.~°
7Insight did not extend to the generally presumed effects of underwriting diversification. To underwrite all lines of property and liability insurance, simply requires, under
New York law, the sum of the capital required for each individual line of insurance. That
is not true, however, of all statutes.
8 State of Illinois Department of Insurance, Financial Regulation of Illinois, December
1977, p. 49.
9 Memorandum of the New York State Insurance Department, February 14, 1977.
1°Estimates of the aggregate capital of the remaining 69 insurers are not conveniently
available. Some of these could have been sizable, however, because of the formation of some
relatively large physicians’ cooperatives in response to market shortages of medical malpractice insurance in the mid-1970s. During this same interval, a few of the large mutual
life insurers also established property-liability subsidiaries, contributing substantial amounts
of capital in the process.
REGULATION OF FINANCIAL INSTITUTIONS
228
TABLE 1
Number of New Property and Liability
Insurers: 1969 to 1978
Year
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
Numb~ ofNew Compan~s
15
32
34
63
58
35
25
46
38
44
390
Source: Best’s Review, Property/Casualty Edition, March 1979, p. 10.
Although the number of new insurers formed in the past 10 years provides
an idea of entry magnitude, some important details are obscured. That interval
includes the formation of some very important entrants. The Prudential Property and Casualty Insurance Company was incorporated in 1972 and held an
initial capital of $1,000,000 and contributed surplus of $4,000,000. Capital was
raised to $2,500,000 in 1973 and in 1974, the parent organization contributed
$44,000,000 more to surplus. By the end of 1976, its net premium volume was
over $300,000,000. The Metropolitan Property and Liability Insurance Company was also formed in 1972, with a paid-in capital of $2,000,000 and contributed surplus of $8,000,000. Its premium volume was $116,000,000 at the
end of 1976. Both insurers received over $200,000,000 in surplus contributions
in 1976 from their parents. Other insurers also formed new subsidiaries during
the 1970s.11
Physicians also formed their own medical malpractice insurers during the
1970s. The Medical Insurance Exchange of California, for example, was formed
in 1975 with an initial capital of $500,000. At the end of that year, its net
premiums written totaled over $5,000,000.12
While the entry of large mutual life insurers and medical associations into
the property and liability insurance business may be viewed in some sense as
unusual and nonrecurring events, it does indicate the impact upon the supply of
insurance which new entrants can have. Moreover, the formation of important
new insurers continues. The recent formation of the New York Insurance Exchange, although comprised of individual underwriting syndicates, had to meet
the entry capital requirements of New York law.la It is viewed by many oblt Best’s Insurance Reports, Property and Casualty Edition, 1977, p. 985.
l~lbid., p. 532B.
lain general, the exchange would be free of traditional forms of rate regulation if its
business were confined to larger commercial account business.
PROPERTY UNDERWRITING CAPITAL
HAMMOND-SHAPIRO
229
servers as a potentially large competitive force in both the domestic and international insurance markets. Other states are also considering appropriate statutes
to permit establishment of such insurance exchanges. Also, a relatively recent
Colorado statute was designed to facilitate the formation of "captive" insurers.14
Any change in U.S. tax laws which would permit premiums to such captives to
be wholly deductible would almost certainly spur legislative review of capital
requirement statutes.15
Assessment of current price or supply problems in some areas of insurance may again direct attention towards new entry as a market response.
Products liability and insurance coverages for municipalities are two current
examples. 16 ~ 1’7
Summary of Existing Statutes
Statutes affecting entry do not focus exclusively on capital. Insurers may be
required to deposit securities with the state, specify plans for conducting business, or certify that management is free of association with insurer failures
within the past five years. Nonetheless, the most common statutory requirement
and probably the most important from a solidity viewpoint is the condition
for licensure that each new insurer possess and maintain prescribed amounts of
capital and surplus. The requirements are sometimes so low as to constitute
only a nominal barrier to entry and provide little support for the solidity objective. In other instances, the prescribed capital amounts are more formidable.~a
14In this context, a captive insurer is a wholly owned subsidiary of a non-insurance
parent firm. The principal and frequently sole purpose of the captive is to insure some or all
of the risks of the parent firm. Most captives are domiciled in Bermuda where entry capital
requirements are quite modest. Observers estimate that about 700 to 800 such captives are
domiciled in Bermuda.
~SThe interest in ease of entry would likely dominate discussions of safety because the
insurers’ policyholder is a corporate parent and not a member of the public. Legislation can
take on several concerns, however, and one is a possible requirement for captives to write
at least some "public" insurance, in which case concerns again focus on solidity.
16Becanse underwriting returns in most lines of insurance are cyclical, an adequate
insurance supply is almost certain to deteriorate. If the deterioration is severe, and if the
line of insurance is major, corrective legislative and/or market forces usually appear.
~TFor instance, Werner Pfennigstorf of the American Bar Foundation observes that
municipalities frequently experience difficulty in obtaining insurance coverage and suggests
the development of a mutual insurance organization by municipalities as a possible solution.
Capital requirements for entry would again be prominent along with the concerns of balance
between solidity and ease of entry. Werner P. Pfennigstorf, "Insurance Mutuals: A Solution
to Municipal Risks Coverage," Risk Management, September 1978, pp. 12-21.
~aSee Allen L. Mayerson, "Enduring the Solvency of Property and Liability Insurance
Companies," Insurance, Government, and Social Policies, Herbert S. Denenberg and Spencer
L. Kimball, eds. (Homewood, Illinois: Richard D. Irwin Co., 1969), pp. 151-160. Mayerson
generally believed capital requirements to be inadequate. See, also, P.L. Joskow, "Cartels,
Competition, and Regulation in the Property-Liability Insurance Industry," The Bell Journal
of Economics and Management Science, Autumn, 1973, pp. 388-391. Joskow generally
assesses all entry barriers to be low, including capital requirements. The only exception is
the relatively higher promotional costs of new direct writers which are incurred to establish
product recognition.
230
REGULATION OF FINANCIAL INSTITUTIONS
Existing statutes typically specify amounts for both initial capital and
initial surplus. The majority of states permit a surplus less than the initial minimum after an insurer has operated for a period of time, five years being common.
For example, the California statute requires an initial capital and surplus of
$200,000 each for entry into fire insurance. After five years, surplus can be
$100,000. The lowered requirement suggests that expenses of established insurers can be met out of operating income and that increasing size (to the
extent associated with increasing age of the insurer) may provide greater stability
in underwriting experience.
Statutes typically differentiate capital requirements by line of insurance and
according to the legal form of insurer. Since mutual insurers issue no capital
stock, they are generally expected to begin with more surplus than stock insurers.
The typical requirement is to specify that the initial surplus of mutual insurers
equal the combined amount of capital and surplus for stock insurers. The same
stipulation generally holds for reciprocal and Lloyds’ organizations. The vast
majority of statutes indicate capital requirements for each separate line of
insurance (e.g., fire) or for general groupings of insurance lines such as "property"
or "casualty." Those insurers which wish the authority to write any or all forms
of nonlife coverages must meet the requirement set by the law for "multipleline" insurance.
Property vs. Casualty Requirements.
Many statutes do not show insurance lines in detail but rather merge fire
and allied coverages into the single designation of "property and liability"
coverages and miscellaneous casualty lines into "casualty." Differences in capital
requirements for different lines or line groupings presumably reflect regulatory
judgments about the relative riskiness of insurance lines: if one line requires
more capital than another, a judgment has been made that it is more risky. The
most consistent distinctions are between the capital required for writing casualty
and that required for writing property (or similarly for writing liability or
writing fire). They are shown in Table 2.
The majority of statutes make no distinction between the capital requirements for property or casualty even though lines in the latter category include
such troublesome coverages as products liability and professional malpractice.
The volatility of these coverages, however, may be too recent to be reflected in
existing laws. Only one statute, however, implies that casualty lines are less risky
(i.e., require less capital) than property lines and several view casualty lines as
riskier than property coverages.
Statutes frequently identify fidelity or surety bonding for separate capital
requirements. Statutes sometimes alter casualty capital requirements depending
on whether or not surety bonds are to be written. If they are, as part of the
casualty license, additional capital is sometimes required. Table 3 summarizes
these distinctions by contrasting, as in Table 2, fidelity and surety requirements
relative to capital required for transacting a property insurance business.
PROPERTY UNDERWRITING CAPITAL
HAMMOND-SHAPIRO
231
TABLE 2
Casualty Capital Requirements as a Percentage
of Property Capital Requirements
Number of States1~
Percentage
0.50
1.00
1.14
1.30
1.33
1.42
1.50
1.67
2.00
1
24
1
1
5
1
5
1
2
Source: Compiled from National Association of Independent Insurers Summary.
TABLE 3
Fidelity or Surety Requirements as
a Percentage of Property Capital Requirements
Percentage
0.50
1.00
1.14
1.25
1.33
1.42
1.50
1.67
2.33
2.40
2.50
Number of Statutes2°
1
17
1
7
2
3
1
1
1
1
2
Source: Computed from N.A.I.I. Summary.
~gThe wording of some statutes makes it difficult to determine capital differentiation
based on "property" or "casualty" or "fire" or "liability." Totals, therefore, amount to less
than 50. These statutes are omitted from the tabulation.
2°See footnote 19.
REGULATION OF FINANCIAL INSTITUTIONS
232
In general, fidelity and surety lines require about the same capital as
casualty, although some states require more and some require less.
Multiple-Line Requirements.
An interesting and important dimension of capital requirement statutes is
the extent to which the diversification effects of multiple-line operation are
recognized. If risk is reduced through the writing of several lines of insurance
instead of only one, then the capital required for writing several lines should
be less than the sum of all individual-line capital requirements. Nearly all statutes
reflect presumed diversification benefits, but to varying degrees. Table 4 shows
multiple-line requirements as a percentage of the sum of individual-line capital
requirements; the lower the percentage, the greater is the recognition of the
presumed benefit of underwriting diversification.
Table 5 shows the absolute requirements for multiple-line licensure.
A plurality of states require about one-third the capital for multiple-line
activity compared to the aggregate of individual-line requirements. The highest
requirement for multiple-line licensure is approximately $6,000,000, while the
lowest is $300,000.
TABLE 4
Multiple-Line Capital Requirements as a Percentage
of the Sums of Capital Requirements for Individual Lines21
Percent of Individual Total
less than 10
10-19
20-29
30-39
4049
50-59
60-69
70-79
80-89
90-99
100
Source: Compiled from N.A.I.I. Summary.
See footnote 19.
Number of States
1
3
3
13
2
5
4
1
3
0
4
PROPERTY UNDERWRITING CAPITAL
HAMMOND- SHAPIR 0
233
TABLE 5
Capital Requirements for Multiple-Line Licensure
Requirements
(000 omitted)
$0- 499
500- 999
1,000-1,499
1,500-1,999
2,000 and over
Number of States
3
10
14
8
3
Source: Compiled from N.A.I.I. Summary.
Judgmental Statutes.
Vermont and Wisconsin allow capital requirements to be set partly at the
discretion of the insurance commissioner. These statutes, particularly the
Wisconsin law, warrant elaboration.
(1) Vermont Statute. The Vermont law for stock insurers reads as follows:
To qualify for authority to transact the business of insurance, a stock insurer
shall possess and thereafter maintain unimpaired paid-in capital of not less
than $250,000 and, when first so authorized, shall possess free surplus of
not less than $150,000.
The Commissioner may prescribe additional surplus if it appears to him that
the kind of insurance to be transacted so requires. (emphasis supplied)~2
The commissioner has the same discretion with respect to mutual insurers.
(2) Wisconsin Statute. The Wisconsin statute is more elaborate and requires
the proposed insurer to file a business plan with the commissioner as well as
allowing the commissioner discretion with respect to initial capital requirements.
It reads as follows:
(1) The commissioner shall specify the minimum capital for a stock corporation or the minimum permanent surplus for a non-assessable mutual being
organized under this chapter. It shall be sufficient, in accordance with sound
business practice, to provide for the needs of the proposed business, but in
no case except a segregated account bearing no risks that are not assumed
by the corporation’s general account shall it be less than $200,000, nor
shall it be more than $2,000,000. In specifying the amount, the commissioner shall take into account all the information in the business plan, the
projection supplied under Section 611.13(2)(k), the general economic
situation, the reinsurance market available to the proposed corporation, and
any other factors relevant to its needs for capital and surplus.
~Quote taken from compilation of statutes by National Association of Independent
Insurers.
234
REGULATION OF FINANCIAL INSTITUTIONS
(2) A corporation organized under this chapter shall have an initial expendable surplus, after payment of all organizational expenses of at least 50% of
the minimum capital or minimum permanent surplus specified under sub.
(1), or such smaller percentage as the commissioner specifies.~3
The Wisconsin law goes on to point out the purpose of the statute governing
initial capital and surplus requirements.
The "minimum capital" (for stock corporations) and "minimum permanent
surplus" (for mutuals) are intended to provide solidity at the time a new
corporation is launched, and for its formative period. The amount needed
depends on what the new company intends to do, and it has to be fixed on
the basis of the information given to the commissioner at the time of
incorporation. It is specified under Section 611.19~
The Wisconsin law is similar to the German Insurance Law which calls for
new insurance firms to submit a "Geschaftsplan" (business plan) which is obliged
to set forth the purpose and organization of the insurer, the territory of its proposed operations, and in particular must state the facts and data intended to
show that its future obligations can be continuously met.~s The business plan
specified in the Wisconsin laws is defined to include: the geographical area in
wtfich business is to be done in the first five years; types of insurance to be
written in the first five years; the proposed marketing methods; to the extent
requested by the commissioner, the proposed methods for the establishment of
premium rates; and a projection of the anticipated operating results of the insurer at the end of the first five years of operation, based on reasonable assumptions of loss experience, premiums, and other income, operating expenses, and
acquisition costs.
Statutory Reference to Premium Volume.
Although the Wisconsin statute allows the Commissioner to consider premium
volume by implication (as part of the business plan) in determining capital
requirements, the vast majority of statutes prescribe capital requirements only as
fixed amounts and without reference to premium volume. The New Mexico taw
however, is an exception. It requires a fixed amount of capital for the initial
authorization to conduct business. Subsequent to authorization, insurers with
multiple-line authority must meet additional minimum requirements based upon
annual premium volume. The capital-premium volume relationships are specified
as follows:
Prem~mVo~me
$ 0 - 10,000,000
10 - 25,000,000
over 25,000,000
23Wisconsin Insurance Laws, Section 611.19.
Wisconsin Insuranc~ Laws, Section 6 23.11.
Wisconsin Insurance Laws, Section 611.13(2)(k).
Capi~l
$300,000
$450,000
$675,000
PROPERTY UNDERWRITING CAPITAL
HAMMOND-SHAPIRO
235
The requirements are in addition to the $300,000 for initial authorization
which is required for any line designation. Insurance writing less than multipleline portfolios must meet similar requirements.~6
Capital Requirements for Entry: Summary and Comment:
Statutes governing capital requirements for new insurers are diverse and
amounts of capital required for a given line of insurance vary widely. Statutes
differ on whether to differentiate requirements by line of insurance and the
extent to which risk diversification is recognized varies widely. Most statutes do
not relate capital requirements to premium volume. This diversity reflects, at the
least, different judgments on matters of underwriting risk and solidity regulation.
It may also reflect a lack of adequate information for the legislative process. The
Wisconsin statute in particular places a considerable burden upon regulatory
judgment.
Estimates of Underwriting Risk Relevant to Entry
A study for the National Association of Insurance Commissioners held that
potential problem insurers could be screened out in part by the establishment of
adequate capital and surplus requirements for new insurers.~7 That analysis also
identified underwriting losses as a major cause of insolvencies in the past decade,
with most of the insolvencies involving smaller insurers.28 A more recent study
by Munch and Smallwood similarly suggests that insolvent insurers are below
average size.z9 Their analysis also showed that insolvencies tended to be more
common among firms writing automobile insurance and less common among
firms writing only commercial lines. Their most robust finding was that insolvencies occur more frequently under inexperienced management,a°
Data and Methodology
Sample Characteristics: Our analysis rests upon the analysis of underwriting
data for nearly the complete universe of property and liability insurers operating
26This requirement is similar to the Great Britain Companies Act of 1967 which also
relates minimum capital to premium income. If premium income does not exceed £250,000,
required capital is £50,000. If premium volume is between £250,000 and £2,500,000, 20
percent of premium income must be held as capital. For premium volume in excess of
£2,500,000, capital must be £500,000 plus 10 percent of premium volume in excess of
£2,500,000.
~MeKinsey & Company, "Strengthening the Surveillance System, Final Report to the
NAIC," April 1974, as reprinted in the NAIC Proceedings, Volume II, 1974, p. 234.
28/bid, p. 253.
29 Patricia Munch and Dennis Smallwood, "Solvency Regulation in the Property/Casualty
Insurance Industry," Paper presented at the National Bureau of Economic Research Conference on Public Regulation, Washington, D.C., December 15-17, 1977, p. 42.
3°1bid, p. 43. New companies, of course, are more likely than established firms to have
inexperienced management.
236
REGULATION OF FINANCIAL INSTITUTIONS
in each line of insurance over the 1972-1975 interval.~1 Specifically, for each
line of insurance the study data consist of (1) net premiums earned, (2) net
premiums written, and (3) losses and expenses incurred. The number of insurers
analyzed for each year, beginning with 1972, was: 979, 998, 1,003, and 1,000.
Individual underwriting results in a line of insurance were omitted if annual net
premiums written were less than $1,000 or if the experience produced a negative
loss or expense ratio. The exclusion of results based on less than $1,000 of
premiums was chosen to eliminate a small number of insurers with unusually
small retentions and which introduced a source of variability into the data
inconsistent with the aims of the study. A negative loss or expense ratio reflects
unusual reinsurance transactions and again injects data variation not a part of
study objectives. The data do not differentiate on the basis of the legal form of
insurers.
Entry Risk Measurement: Capital required for underwriting is directly related to the level of risk associated with the activity undertaken. Underwriting
risk, in turn, is d!rectly affected by fluctuations in claim frequency and severity,
competition, economic conditions, insurer size,a2 access to the reinsurance
market, and any regulatory constraints which might be present. The study of
underwriting risk for new insurers is a difficult task which is complicated by the
absence of an operating history.
One approach to the study of new insurer capital needs would be to trace
the experience of a cohort of new entrants in each line of insurance over a period
of time. The problems of changing economic conditions, competition, regulatory
philosphies,aa changing insurer size, and differences in insurance lines entered,~
make such an approach difficult. Moreover, shifts in societal attitudes can quickly
affect both claim frequency and severity. This appears to have been true in
various liability lines, particularly products and medical malpractice. These are all
familiar problems in time series analysis. Therefore, an industry cross-sectional
approach to analysis of the data was chosen, where small premium volumes of
existing insurers were used to estimate the underwriting risk for new insurers,as
31 The data were prepared on tape by A.M. Best and Company for the research purposes
of the study.
a2The pooling or large number effects commonly associated with size tend to reduce
year-to-year variation in those events which are poolable, such as random fluctuations in the
collective value of claims. It does not have the potential to pool out variation resulting from
social change, competition, regulatory constraints and the like.
3a Successive commissioners in a given state may pursue different regulatory philosophies,
and differences among commissioners may also exist at any point in time.
a4Some lines of insurance are not commonly represented in the many cohorts of entrants.
~SA time-series analysis of the problem, utilizing only the experience of new insurers,
would still be helpful. Our approach is not in lieu of that but represents a more convenient
starting point for an unresearched area. Observations of the growth patterns for new insurers
point strongly to the presence of small premium volume, at least in the first year of operation and sometimes longer. Exceptions have been among physicians’ cooperatives entering
the medical malpractice line and insurers formed by large corporate parents and provided
with an initially large capital, sometimes augmented by transfers soon after formation, and
large enough to support a premium volume beyond that usually associated with new insurers.
PROPERTY UNDERWRITING CAPITAL
HAMMOND-SHAPIRO
237
By including 1972, a year of record underwriting profits, and 1975, a year of
record underwriting losses, a focus was provided on well-defined conditions of
strong interest to insurance regulators. The nature of underwriting risk under
those conditions is described, so that they may serve as guidelines should they
again prevail.
The empirical approach is also contrained by the form in which data are
conveniently available. Claims and expense data for each line of insurance are
reported directly to insurance commissioners on the expense exhibit, a prescribed
regulatory form. Data on the actual distribution of individual claim amounts are
available only from internal records of insurers and may not be available at all.36
Therefore, claims and expense data reported on the Insurance Expense Exhibit
were employed. These data can easily be reduced to two values widely used in
the analysis of underwriting gains and losses. They are:
(1) Loss Ratio: The ratio of losses and loss adjustment expenses incurred
to net premiums earned.
(2) Expense Ratio: The ratio of underwriting (nonclaim) expenses incurred to net premiums written.
It is standard practice to add the loss and expense ratios to produce a single
measure of underwriting performance known simply as the combined ratio.37
For the purpose of this study, underwriting return is measured by the complement of the combined ratio, i.e., 1 - (loss ratio + expense ratio) and risk is
measured by the standard deviation of this function. The analysis revolves about
the development of the standard deviation of returns to a particular line of
insurance across specified size subsets of insurers.
Insurance line designations are those which are required for the Insurance
Expense Exhibit. The lines are:
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
Fire
Allied Lines
Farm
Homeowners Multiple-peril
Commercial Multiple-peril
Ocean Marine
Inland Marine
Earthquake
Group Accident and Health
Other Accident and Health
a6 Statutory accounting requirements govern much of the way in which data are mainrained and may discourage the transcription and retention of data in alternate forms.
3~Statutory accounting is a mixture of cash and accrual accounting. Prepaid expenses
are not allowed as an asset. Where premium volume is expanding, underwriting gains are
slightly understated by the statutory system. Combining the loss and expense ratio is a
shorthand approach to estimating underwriting results, taking prepaid expenses into consideration. A combined ratio of 100 percent, for example, means that the underwriting
functions (ignoring possible investment gains or losses) essentially broke even.
238
REGULATION OF FINANCIAL INSTITUTIONS
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
Workers Compensation
Liability Other than Auto
Private Passenger Auto Liability
Commercial Auto Liability
Private Passenger Auto Physical Damage
Commercial Auto Physical Damage
Aircraft
Fidelity
Surety
Glass
Burglary and Theft
Boiler and Machinery
Credit
International
Reinsurance
Miscellaneous
All or Total
Medical Malpracticeas
Diverse coverages, such as accounts receivable insurance, mobile home
insurance, extra expense insurance, "floater" contracts, and countless others are
all submerged among the categories listed above. Therefore, a particular line
designation is not necessarily homogeneous across insurers because the offering
of such coverages is not consistent across all insurers. Thus, a "line" of insurance
may consist of several related coverages.
The cross-sectional measure of risk serves as a good estimate of the underwriting variability for small premium volumes under certain ideal condtions:
(1) The content of a line of insurance is essentially the same for all
insurers writing that line. For example, nonautomobile liability
written by insurer A encompasses the same period and coverages as
for insurer B, C, D, E, etc.a9
(2) All insurers in a given line have essentially the same underwriting
objectives. Variation in underwriting results across insurers represents random departures from a consistent set of objectives and
expectations.
(3) Each insurer faces the same regulatory constraints where underwriting
variations across insurers are independent of regipnal differences that
may reflect different regulatory patterns.
Insurers may, of course, exhibit different underwriting objectives, attempt
to appeal to different markets, write diverse coverages, and operate under vary38Medical malpractice was added to the Expense Exhibit as a separate line in 1975.
Prior to that, it was a part of Liability Other than Auto.
39The line "liability other than auto", for example, is identified by the expense exhibit
as a separate line. However, it encompasses several different liability lines including products
liability and, until 1975, medical malpractice.
PROPERTY UNDERWRITING CAPITAL
HAMMOND-SHAPIRO
239
ing regulatory constraints. Unfortunately, the nature of available data does not
permit consistent identification of such differences among insurers. The results
of the analysis, therefore, can approximate underwriting risk and related values
but not measure it precisely.4°
Underwriting Risk Differences among Lines: The Empirical Evidence
The data base of the present study permits direct cross-sectional observations
of underwriting risk differences among lines of insurance and among any desired
size strata of insurers.
New insurers typically begin operations in one or only a few lines of insurance and premium volume, particularly for a single line, frequently remains
below $500,000 for the first year of operation.41 Accordingly, the data which
follow include annual retained premium volumes in each insurance line up to
$500,000. The size classifications constructed for such analysis are:
$ 1,000- $ 99,999
100,000 - 499,999
By presenting underwriting risk differences only for small size classifications,
the possible effects of size from the inclusion of larger volumes is greatly reduced.4~ Tables 6 through 9 array underwriting risk differences among lines of
insurance for each of the years 1972 through 1975; those lines with lowest risk
values begin each array for each year.
This section attempts to identify from the underwriting risk array of Tables
6 through 9 those lines which presented high levels of underwriting risk in each
of the four years. The identification problem is somewhat analagous to that of
legislators or regulators confronting the same problem. If a line is in fact low risk
but is misclassified as high risk, the error is not in conflict with the solvency or
solidity regulatory objective. To classify a line as low risk, however, when it is
not, does conflict with that objective. Identification, therefore, focuses only on
the high-risk classification. Those lines with the highest risk values for the two
lowest size classifications are identified for each year. Any such lines are then
surveyed for consistency of the high-risk classification across the four year
interval.
4o Some insurers do not even maintain detailed accounts of claim experience for coverages not identified on the expense exhibit. However, if the study were conducted for only
the business written within a particular state, the problem of varying regulatory patterns
would not exist. It would be relatively easy for state departments of insurance to undertake.
41 Observation of the total premium volumes of a sample of 80 new insurers generally
confirms that premium volume usually remains below $500,000 through the first year of
operation and frequently through the first two years. Where volume is in excess of that,
it appears to be clearly associated with capital contributions from a parent corporation.
4~The study data consistently show decreasing risk values to be associated with increasing size classifications. The complete set of data is not shown here but is available
from the authors upon request. A paper on underwriting-size effects is being prepared
separately.
240
REGULATION OF FINANCIAL INSTITUTIONS
TABLE 6
19’72 Variability of Combined Ratios of Insurance Lines: Arrayed by
Standard Deviation: Number of Companies Shown in Parentheses
Net Premium Volume
$1,000 - 99,999
Earthquake
2. Glass
3. Group A & H
4. Commercial Muir. Peril
5. Allied Lines
6. Pvt. Pass. Auto. Phys. Dmge.
7. Credit
8. Farm
9. Burglary and Theft
10.
Ocean Marine
11.
12.
Commercial Auto.
Phys. Dmge.
Fire
13.
Other Accident and Health
14.
Homeowners Muir. Peril
15.
Workers Comp.
16.
Inland Marine
17.
Air
18.
Fidelity
19.
Boiler and Machinery
20.
Liab.
21.
Liab. Other Than Auto.
22.
Pvt. Pass. Auto. Liab.
23.
Surety
24. All Lines
0.0000
( 0)
$100,000 -- 499,999
Earthquake
0.2520
(288)
0.3250
( 45)
0.3580
(lS9)
0.3660
(199)
0.3800
Burglary and Theft
( 70)
Homeowners Mult. Peril
0.3810
(47)
0.4000
Pvt. Pass. Auto. Phys. Dmge.
0.4140
(277)
0.4220
Commercial Auto.
Phys. Dmge.
Allied Lines
0.4330
(176)
0.4400
(102)
0.4750
(66)
0.6450
( 72)
0.8230
(60)
0.8270
(284)
0.8370
(54)
0.9770
(104)
1.1590
( 58)
1.6070
( 9S)
2.5890
(177)
2.6890
( 51)
5.6310
(73)
0.9660
(27)
Fidelty
( 11)
(
Glass
Farm
Fire
Inland Marine
Air
Credit
Other Accident and Health
Commercial Muir. Peril
Commercial Auto. Liab.
Group A and H
Ocean Marine
Liab. Other Than Auto.
Workers Comp.
Pvt. Pass. Auto. Liab.
Boiler and Machinery
Surety
All Lines
0.0000
(o)
0.1510
( 72)
0.1550
(47)
0.1830
( 8)
0.2180
(211)
0.2200
(145)
0.2270
(129)
0.2290
(167)
0.2290
(145)
0.2390
(211)
0.2650
(27)
0.2770
(26)
0.3110
(27)
0.3230
(45)
0.3310
(113)
0.3390
(12S)
0.4300
( 38)
0.4650
( 39)
0.4730
(177)
0.4790
(88)
0.6060
( 77)
0.6770
( 13)
1.4150
( S8)
0.3450
( 69)
~ROPERTY UNDERWRITING CAPITAL
HAMMOND-SHAPIRO
241
TABLE 7
1973 Variability of Combined Ratios of Insurance Lines: Arrayed by
Standard Deviation: Number of Companies Shown in Parentheses
Net Premium Volume
$1,000 - 99,999
1. Group A & H
2. Credit
3. Glass
4. Commercial Multi. Peril
5. Inland Marine
6. Commercial Auto.
Phys. Dmge.
7. Pvt. Pass. Auto. Phys. Dmge.
8. Fire
9. Burglary and Theft
10.
Liab. Other Than Auto
11.
Commercial Auto. Liab.
12. Other Accident and Health
13.
Farm
14. Allied Lines
15.
Earthquake
16. Surety
17.
Ocean Marine
18.
Fidelity
19.
Homeowners Multi. Peril
0.1800
(40)
0.2450
(42)
0.3100
(298)
0.3900
(147)
0.4010
(271)
0.4560
(178)
0.5330
0.5840
(107)
0.6210
(264)
0.6270
(166)
0.6540
(100)
0.6770
( 64)
0.7150
( 90)
0.7490
(205)
0.7840
( 60)
0.8460
(76)
0.8850
( 70)
0.9000
(97)
0.9130
(92)
$100,000 -- 499,999
Burglary and Theft
Other Accident and Health
Glass
Boiler and Machinery
Farm
0.2290
Pvt. Pass. Auto. Phys. Dmge.
0.2310
Allied Lines
0.2470
(211)
0.2480
(142)
0.2620
(133)
0.2660
(19S)
0.2730
( 84)
0,2970
(155)
0.3160
(74)
0.3580
Homeowners Multi. Peril
Commercial Multi. Peril
Inland Marine
Workers Comp.
Commercial Auto.
Phys. Dmge.
Pvt, Pass. Auto, Liab,
Credit
Fidelity
Commercial Auto. Liab.
Liab. Other Than Auto.
Air
Group A and H
20. Workers Comp.
1.1220
Earth qua ke
21.
Air
Ocean Marine
22.
Pvt. Pass. Auto. Liab.
23.
Boiler and Machinery
24.
All Lines
1.2560
( 58)
1.3120
( 64)
2.4770
(49)
0.6350
( 22)
(70)
0.1830
(79)
0.1970
( 39)
0.2100
(43)
0.2270
( 16)
Surety
Fire
All Lines
( 52)
(128)
( 17)
0.3870
(30)
0.3910
(143)
0.4140
(179)
0.4280
(41)
0.4350
(41)
0.4540
(23)
0.5150
(42)
0.9730
( 57)
3.5660
(226)
5.4380
(56)
242
REGULATION OF FINANCIAL INSTITUTIONS
TABLE 8
1974 Variability of Combined Ratios of Insurance Lines: Arrayed by
Standard Deviation: Number of Companies Shown in Parentheses
Net Premium Volume
$1,000 -- 99,999
1. Glass
2. Earthquake
3. Other Accident and Health
4. Ocean Marine
5. Pvt. Pass. Auto. Phys. Dmge.
6. Burglary and Theft
7. Allied Lines
8. Farm
9. Fire
10.
Air
11.
Commercial Multi. Peril
12.
Fidelity
13.
Surety
14.
Commercial Auto. Liab.
15.
16.
Commercial Auto.
Phys. Dmge.
Pvt. Pass. Auto. Liab.
17.
Group A and H
18.
Inland Marine
19.
Homeowners Multi. Peril
20.
Credit
21. Workers Comp.
22.
Boiler and Machinery
23.
Liability Other Than Auto
24.
All Lines
0.2800
(301)
0.3760
(60)
0.4100
( 65)
0.4670
( 65)
0.4980
( 87)
0.5910
(270)
0.6010
(216)
0.6500
(82)
0.6700
(116)
0.7740
( 48)
0.7830
(149)
0.8370
(109)
0.9760
(86)
1.0730
(121)
1.0930
(193)
1.1540
(70)
1.1960
(46)
1.2810
(237)
1.4330
(101)
2.3000
(42)
2.3670
(88)
2.4910
( 53)
13.6530
(167)
2.5980
17)
$100,000 - 499,999
Group A and H
Glass
Other Accident and Health
Earthquake
Pvt. Pass. Auto. Phys. Dmge.
Burglary and Theft
Commercial Multi. Peril
Fire
Farm
Commercial Auto.
Phys. Dmge.
Allied Lines
Inland Marine
Homeowners Multi. Peril
Ocean Marine
Boiler and Machinery
Fidelity
Workers Comp.
Pvt. Pass. Auto. Liab.
Commercial Auto. Liab.
Liability Other Than Auto
Air
Surety
Credit
All Lines
0.159(
( 41~
0.177(
( 371
0.216(
( 43’,
0.232C
( 28’,
0.27912
( 130’,
0.2961~
( 81~
0.2980
( 126~
0.2990
( 2241
0.3040
( 61)
0.3140
(148)
0.3380
(214)
0.3580
(192)
0.3620
(135)
0.4570
( 61)
0.4640
( 22)
0.4800
( 32)
0.4910
( 72)
0.4910
(73)
0.5210
(142)
0.5900
(186)
0.8260
(46)
0.9080
(62)
0.9200
(24)
0.3150
(61)
’ROPERTY UNDERWRITING CAPITAL
HAMMOND-SHAPIRO
243
TABLE 9
1975 Variability of Combined Ratios of Insurance Lines: Arrayed by
Standard Deviation: Number of Companies Shown in Parentheses
Net Premium Volume
$1,000 - 99,999
Earthquake
2. Glass
3. Credit
4. Fidelity
5. Allied Lines
6. Homeowners Multi. Peril
7. Farm
8. Commercial Auto. Liab.
9. Burlary and Theft
10.
Pvt. Pass. Auto. Phys. Dmge.
11.
Group A and H
12.
Liability Other Than Auto
13.
Air
14.
15.
Commercial Auto.
Phys, Dmge.
Inland Marine
16.
Pvt. Pass. Auto. Liab.
17.
Commercial Multi. Peril
18.
Surety
19.
Fire
20.
Ocean Marine
21.
Other Accident and Health
22.
Boiler and Macheriny
23.
Medical Malpractice
24.
Workers Comp.
25.
All Lines
$100,000 -- 499,999
0.2770
(70)
0.3320
(286)
0.3800
(34)
0.4520
(99)
0.4700
(209)
0.5070
( 87)
0.5440
(86)
0.5600
(106)
0.6120
(268)
0.6160
(74)
0.7740
(46)
0.8060
(167)
0.8560
(44)
0.8690
(187)
0.8770
(244)
0.8780
(4"/)
1.0530
(112)
1.1000
(86)
1.1300
(115)
1.2900
( 55)
1.3050
(70)
1.6360
( 53)
5.3900
(22)
9.3000
(6O)
0.3400
( 10)
Glass
Farm
Fire
Earthquake
Commercial Auto.
Phys. Dmge.
Other Accident and Health
Pvt. Pass. Auto. Phys. Dmge.
Allied Lines
Boiler and Machinery
Homeowners Multi. Peril
Burglary and Theft
Inland Marine
Commercial Auto. Liab.
Group A and H
Air
Liability Other Than Auto.
Ocean Marine
Pvt. Pass. Auto. Liab.
Workers Comp.
Fidelity
Surety
Medical Malpractice
Credit
Commercial Multi Peril
All Lines
0.1870
( 35)
0.2920
( 63)
0.2970
(215)
0.2970
( 2S)
0.3130
(157)
0.3150
(40)
0.3200
(135)
0.3310
(213)
0.3320
(26)
0.3490
(135)
0.4080
( 78)
0.4190
(186)
0.4630
(134)
0.4770
(36)
0.5410
( 57)
0.5530
(168)
0.5530
( 52)
0.6360
( 84)
0.6830
( 76)
0.7240
(42)
0.8000
( 50)
0.8100
( 33)
0.8820
(24)
1.2430
(151)
0.9"/30
(69)
244
REGULATION OF FINANCIAL INSTITUTION~
There are 24 principal lines of insurance identified on the expense exhibit.4~
The eight lines exhibiting the highest underwriting risk values were classified as
high or possibly high risk~ depending upon whether they appear among the tol:
eight in both of the size categories. Specifically, the judgmental criteria used fo~
such identification were as follows:
(1) If a line is among the eight most risky lines in both of the size classifications, it is classified as high risk.
(2) If a line is among the eight most risky lines in either of the smalles~
size classifications, it is classified as possibly high risk.
The classification of a line as either high risk or possibly high risk in a given
year may not be significant. Assume, for example, that in any given year each
line has an equal opportunity to fall within the high-risk category. Under that
condition, the probability that a line would be in the high-risk grouping would
be 1/9 or 0.111. The probability that a line would be in the possibly high-risk
category is 4/9 or 0.444~4 The appearance of any line in a single year as higl~
risk or possibly high risk need not, therefore, be surprising.
The classifications, of course, apply only to the years under observation..
Table 10 identifies those lines judged as high risk or possibly high risk.
If a line remained in the high-risk classification over the entire four-year
interval, that would offer improved affirmation of its high-risk status. Also, if a.
line appeared as either high risk or possibly high risk in each of the four years,,
that too would be further affirmation of its high-risk potential. These kinds of
observations are summarized in Table. 11.
As indicated previously, if it is assumed that each line has an equal chance
of being identified as high or possibly high risk in a given year, the probability i
relatively high that it will occupy one of these classifications. Should it remain
in either of those groupings for every year of the study, however, it would lend
considerable credence to the hypothesis that a given line is, in fact, risky. Again,
assume that the underwriting risk value of a given line is a random variable,
independent and identically distributed with equal probability of being in any
risk category. Under these conditions, the chance that a particular line will
remain in the high-risk category over the four years is (1/9)4 or 0.00015; for
three of the four years 32/94 or 0.0049; and for two of the four years 384/94 or
0.05853.4s Similarly, the probability of being in either but not both, the high-
43 International, miscellaneous, and reinsurance are not included in the presentation.
Although they are a part of the data, they do not represent well-defined loss exposures and
have little, if any, relevance to capital statutes dealing with entry.
4~The probability of falling into the high-risk grouping is 1/3 for each of the two smallsize categories, so that the probability of the joint occurrence is 1/3 x 1/3 = 1/9. The
probability of falling in only the possibly high-risk grouping is 2/9 + 2/9 - 0 = 4/9.
4SThe probability that a line will be a high-risk line for n of the four years is:
(4n) (_~)n (_~)
TABLE 11)
Lines Developing High Cross-Sectional Risk Values: 1972-1975
!972
High Risk
-Surety
-Private Passenger Automobile Liability
-Liability
Other Than
Automobile
-Commerc~l
Automobile
Liability
-Boiler and
Machinery
!973
Possibly
High Risk
High Risk
1974
Possibly
High Risk
High Risk
1975
Po ss~ly
High Risk
High Risk
Possibly
High Risk
-Bo~qer and
-Liability
-Workers Corn- -Boiler and
Other Than
Machinery
pensafion
Machinery
-Air
-Ocean Marine
Automobile -Homeowners -Medical Mal- -Other Accident
-Private
Pas-Inland Marine -Surety
senger Auto- -Workers Corn- -Irdand Marine practice
and Health
-Workers Cornmobile
pensafion
-Ocean
Marine
-Fke
pensation
-Group Acci-Workers Corn- -Credit
-Surety
-Private Pasdent and
-Ocean Marine
pensation
-Private Passenger AutoHealth
-Commercial
-Group Accident
-Homeowners
senter Automobile
Multipte
-Fidelity
and Health
mobile
Liability
-Fidelity
Peril
L~ability
-Commercial
-Credit
-Commercial
Automobile
Automobile
L~ability
Liability
-Air
-Liability
-Surety
Other Than
Automobile
-Group Accident
and Health
-Earthquake
-Fidelity
-Air
~Boiler and
Machinery
-Fire
TABLE 1 !
High Risk Classification Consistency Across
1972-1975 Interval for Small Premium Volumes
Number of Years Appearing as Either High Risk or
Possibly High Risk
Number of Years Appearing as High Risk
ONE
-Commercial
Automobile
Liability
-Boiler and
Machinery
-Air
-Credit
-Medical Malpractice
-Commercial
Multiple
Peril
TWO
THREE
-Private Pas-Surety
senger Automobile
Liability
-Llabiilty
Other Than
Automobile
-Other
-Ocean Marine
-Workers Compensafion
FOUR
-
ONE
-Earthquake
TWO
THREE
-Inland Marine -Fideliry
FOUR
-Workers Compensation
-Fire
-Air
-Private
Pas-Credit
-Ocean Marine
senger
Auto-Commercial -Homeowners -Group Accimobile
Multiple
Multiple
dent and
Liabifity
Peril
Peril
Health
-Surety
-Boiler and
Accident
Machinery
and Health
-Commercial
Automobile
Liability
-Liability
Other Than
Automobile
-Medical Malpractice
’ROPERTY UNDERWRITING CAPITAL
HAMMOND-SHAPIRO
247
isk and the possibly high-risk classification in all four years is 0.0951 and in
hree of the four years is 0.2024.46
Three lines appeared in either the high- or possibly high-risk categories over
he four years: workers compensation, private passenger auto liability, and
;urety. While no line appeared as high risk for four years, surety appeared for
:hree, thereby strongly suggesting it to be, in fact, high risk. Private passenger
into liability, liability other than auto, workers compensation, and ocean marine
~ppeared as high risk for two years.47
Risk classification was also performed for premium volume up $1,000,000
~y adding a third size class of $500,000 - $999,999. The number of lines identii’ied as high. or possibly high-risk was fewer. However, surety again continued in
9ne of these categories in at least one of the three size groupings in each of the
l’our years. Similarly, liability other than auto continued for three years.
Cross-sectional risk values for small premium volumes provide a basis for
developing hypotheses and supplementing regulatory judgment regarding initial
capital requirements among lines. The data may be especially useful to Wisconsintype statutes which allow regulatory discretion in establishment of initial capital
amounts and for revisions of statutes specifying fixed capital amounts for entry
into various insurance lines.
While the classification scheme for determining high-risk and potentially
high-risk classes may also be used to identify low.risk counterparts, it is important to note again the consequences of misclassification error. If a line is classified
as high risk when it is not, the consequence is over-capitalization, at least initially.
On the other hand, misclassification of a line as low risk could lead to insolvency.
Until more data are available, prudence is required in the identification of lowrisk lines from the study data.
Some lines did appear to be low risk, using the same classification system as
before. Glass appeared in the low-risk classification in each of the four years
and private passenger automobile physical damage appeared as low risk in three
of the four years. Other lines did not develop such consistency. The high-risk
classifications of Tables 10 and 11 suggest that the highest capital requirements
for entry should be for surety and perhaps for private passenger automobile
liability, liability other than automobile (principally products and malpractice),
ocean marine, and workers compensation. These lines developed the most consistent high-risk patterns.48
46 The probability that a line will be in either, but not both, the high-risk and the
possibly high-risk classifications in n of the four years is:
n~ ! n~ ! (4 - n~ - n2)!
nI + n: = n
47Liability other than auto would almost certainly have appeared as high risk for three
years had not the medical malpractice component of the line been separately identified for
the last year, 1975. The latter appeared as high risk for that year.
~8 The experience of the liability other than auto line may change given it no longer
includes medical malpractice; It does encompass products liability, however, and that
generally remains as a troublesome line.
248
REGULATION OF FINANCIAL INSTITUTIONS
As noted, automobile physical damage produced lower risk patterns than
automobile liability. Statutes specifying entry capital requirements, however,
frequently make no such distinction. Statutes do, however, frequently require
.high capital for entry into the surety lin~s, a requirement consistent with the
empirical observations shown here. Since medical malpractice was identified as
a separate line only in 1975, it could have been separately identified only once.
Its 1975 risk value and the situation leading to price and supply problems
strongly suggest that classification as a high-risk line would be reasonable.
Low-risk lines tend to be those with relatively low catastrophe potential.
Both the glass and automobile physical damage risks, for example, are characterized by a well-defined and relatively low maximum loss per insured unit
and a relatively small chance of a single occurrence effecting large numbers of
exposures simultaneously. On the other hand, risky lines tend to have a higher
catastrophic potential. Surety contracts cover diverse loss exposures and aggregate claims can be unpredictable and large. Changing socio-economic conditions
and attitudes have elevated the large-loss potential of all liability lines. It is not
surprising that private passenger automobile liability and liability other than
automobile fall into high-risk groupings (with medical malpractice as well).49
The workers compensation risk is beset with similar pressures, higher medical
care costs and income payments increased by inflationary pressures. Ocear
marine is also characterized by a large maximum loss potential.
Capital and Premium Volume
Statutory capital requirements typically make no reference to new insure~
premium volumes. Yet, the adequacy of any capital amount can be judged onb
in reference to the premium volume which may be associated with it. Premiums,
are approximately the expected value of future losses and expenses and if actuai
underwriting results always coincided with those expected, there would be nc
need for a financial cushion to absorb fluctuations in underwriting experience
In the context of underwriting, capital is equivalent to a financial buffer.
The presumed stabilizing effects of large premium volumes suggest that ne~
insurers, to remain at a given level of safety, would be constrained to a smalle
ratio of premiums to capital than larger firms. The ratio is not only conceptuall,.
important but also continues as one of the key guidelines used by regulators i~
assessing solvency. Over the years, a premiums-to-capital ratio of 2.0 has been
commonly used rule-of-thumb in judging the capital adequacy of all nonlif~
insurers,s° The ratio, however, has usually been applied and spoken of withou
reference to insurer size or to differences in the mix of underwriting portfolio.
49The finding is consistent with the Munch and Smallwood findings that insolvenci,
tended to be greatest among firms writing automobile insurance. Munch and Smallwoo,
p. 42.
S°The 2.0 ratio is usually referred to as the "Kenney Rule": that an insurer’s premiu:
volume should not normally exceed 200 percent of its capital and surplus. The Early Warni~
System of the National Association of Insurance Commissions prescribes 3.0 as an upp
limit but still without direct reference to size or business mix.
PROPERTY UNDERWRITING CAPITAL
HAMMOND-SHAPIRO
249
The study data can be employed to estimate maximum ratios of premiums
to capital for each line of insurance and for any size classification of insurers.
The most straight-forward estimation can be made from a standard statistical
formulation resting upon the assumption of normality in the underlying data,
in this case, the distribution of the combined loss-and-expense ratio.
Specifically:
where:
C = P(Z" ox+X- 1)
C = capital
X = expected combined ratio
P = estimated premium volume
ox = estimated standard deviation of expected combined ratio
Z = value from a normal distribution associated with a selected
probability of ruin value.
The formulation was employed in this context a decade ago by Hoffiander.sl
His study provided a major empirical insight into the relationship between
monoline underwriting variability and maximum ratios of premium to capital.
For purposes of illustration, it is assumed that the highest probability of
ruin acceptable to regulators is 0.001 (Z = 3.09). For example, if the data for a
given line indicate an average combined ratio of 0.99 with a standard deviation
of 0.05, the maximum premiums-to-surplus ratio for that line would be estimated
as follows:
C = 100 (3.09 x 0.05 - 0.01)
C = $14.45 for every $100 of premiums or an estimated permissible
ratio of 6.92.
In the above illustration, the combination of profitable and reasonably
stable underwriting results produced a relatively high ratio of premiums to
surplus.
Table 12 shows the maximum premiums to capital ratios that an insurer
could have written, under the assumptions noted, in only one line of insurance
and not have exceeded a ruin probability of 0.001. The values, therefore, do not
reflect possible balancing effects from writing more than a single line. Only in
the values shown for all lines combined is there any indication of the direction
and extent of possible diversification effects.
Size difference observations from Table 12 are confined to the two smallest
premium volume classes ($1,000 to $99,999 and $100,000 to $499,999) and to
sa Alfred E. Hofflander, "Minimum Capital and Surplus Requirements for Multiple Line
Insurance Companies: A New Approach," Insurance, Government, and Social Policy."
Studies in Insurance Regulations, edited by Spencer L. Kimball and Herbert S. Denenberg,
S.S. Huebner Foundation for Insurance Education (Homewood, Illinois: Richard D. Irwin
Co., 1969). See pp. 80-88 in particular. Hofflander, however, did not attempt to differentiate the ratios on the basis of different size classifications. His work did draw attention to
the problem of using arbitrary ratios to assess insurer solidity and capital adequacy.
TABLE 12
Maximum Premiums-to-Capital Ratios for Monoline Underwriting and Selected LLne Groupings;
For Small Premium Volume and All Insurers Combined, 1972-1975
1973
1972
Line
$1,000- $100,00099,999 499,999 All
1974
1975
$1,000- $100,000$1,000- $100,00099,999 499,999 All
99,999 499,999 All
$1,000- $100,00099,999 499,999 All
1. Fire
0.79
1.84
8.68
0.55
0.09
0.61
0.46
1.06
2.73
0.28
1.11
2.65
2. A!lied Lines
1.12
2.23
5.25
0.46
1.86
3.57
0.53
1.00
1.73
0.71
1.06
2.15
3. Farm
0.80
2.35
3.10
0.44
1.41
2.09
0.44
0.94
1.31
0.55
0.96
1.49
4. Homeowners
Peril
0.48
1.57
5.29
0.33
1.28
4.79
0.20
0.78
2.07
0.55
0.80
2.37
5. Com. Mult. Peril
1.14
1.17
6.60
0.97
1.63
5.04
0.42
1.24
2.58
0.31
0.26
1.80
6. Ocean Marine
0.84
0.69
2.57
0.34
0.59
2.15
0.73
0.66
1.99
0.24
0.51
1.40
7. Inland Marine
0A1
1.68
2.47
0.94
1.49
3.26
0.25
0.94
1.96
0.36
0.77
2.12
8. Earthquake
NA
NA
NA
0A8
1.03
0.86
1.48
3.57
4.39
2.97
2.21
5.78
9. Group A & H
1.04
0.76
3.05
1.91
0.72
2.72
0.26
2.21
3.00
0.40
0.66
2.57
10. Other A & H
0.66
1.12
2.77
0.47
1.99
2.41
0.76
1.74
2.08
0.25
1.05
1.82
11. Workers Comp.
0.37
0.69
3.35
0.27
1.28
3.13
0.13
0.62
2.41
0.03
0.45
1.41
12. LLab. Other
Than Auto
0.12
0.71
1.09
0.52
0.78
1.32
0.02
0.52
0.43
0.40
0.56
1.02
13. Pvt. Pas. Auto.
Liab.
0.11
0.49
3.43
0.22
0.93
3.32
0.26
0.59
2.90
0.33
0.46
2.74
14. Com. Auto. Liab. 0.20
1.03
1.91
0AT
0.80
1.97
0.28
0.61
1.58
0.56
0.67
1.49
15. Pvt. Pas. Auto.
P.D.
0.80
1.59
6.93
0.59
1.36
4.01
16. Com. Auto. P.D.
0.78
1.70
4.15
0.72
1.10
17. Air
0.38
1.21
1.63
0.22
0.64
18. Fidefity
0.32
1.55
1.81
0.35
0.92
19. Surety
0.06
0.22
0.69
0.37
0.32
3.27
0.81
1.63
1.35
20. Glass
1.41
2.35
2.04
1.09
1.64
21. Burg. & Theft
0.88
4.04
3.79
0.54
2.62
22. Boiler & Mach.
0.24
0.50
1.73
0.13
1.81
0.85
0.88
1.10
0.54
1.32
24. Med. Malpractice
NA
NA
NA
NA
NA
25. All
0.31
0.97
7.20
0.46
0.06
26. !+2
1.03
1.86
12.14
0.54
0.10
0.08
23. Credit
27. 1+2+3+4
0.46
1.39
10.90
0.42
28. 13+!5
0.24
0.95
5.51
0.30
1.07
29. 14+16
0.21
1.03
2.52
0.67
0.89
30. 13+14+15+16
0.13
1.14
5.07
0.39
0.98
0.08
31. 1+2+3+4+22
0.46
1.39
10.90
0.42
32. 1+2+3+4+12
0.12
0.93
3.57
0.45
0.08
33. 1+2+3+4+5
0.47
1.31
13.93
0.46
0.08
34. 11+12+13+14
0.06
0.47
2.64
0.45
0.88
1.85
2.50
1.38
0.16
NA
2.68
0.89
1.29
4.18
2.57
4.22
1.30
1.32
!.54
3.22
0.59
0.29
0.35
0.37
0.33
1.09
0.56
0.12
0.13
NA
0.!2
0.41
0.47
0.32
0.42
0.79
0.47
0.27
0.38
0.53
!.07
1.01
0.33
0.65
0.34
1.62
1.19
0.68
0.30
NA
0.92
1.19
0.87
0.94
0.62
0.87
0.87
0.52
0.84
0.59
2.01
1.76
0.78
1.24
0.66
1.69
1.77
1.!4
0.16
NA
3.20
3.11
2.84
2.90
1.95
3.38
2.86
1.76
3.20
2.41
0.43
0.36
0.34
0.72
0.29
0.90
0.52
0.19
0.96
0.05
0.78
0.24
0.20
0.31
0.54
0.25
0.20
0.23
0.21
0.49
0.81
0.96
0.49
0.42
0.38
1.52
0.81
0.97
0.35
0.36
0.30
1.06
0.80
0.91
0.75
0.76
0.80
0.71
0.71
0.90
2.03
1.81
0.76
0.92
0.64
1.32
1.53
1.60
0.59
0.55
3.15
3.06
2.95
2.33
1.89
2.49
2.97
2.28
3.23
2.38
252
REGULATION OF FINANCIAL INSTITUTIONS
facilitate comparisons with other insurers, the ratio value calculated from the
results of all insurers combined. Size effects are readily apparent. The underwriting experience at small premium volumes consistently develops maximum
ratios below those calculated from the experience of all insurers combined.
Volumes in excess of $500,000 are consistent with the pattern shown.
Table 12 also shows maximum ratios of premiums to capital for certain
selected but standard groupings of insurance lines. The values shown were calculated on the basis of the previously noted assumptions reflecting a ruin probability of 0.001. The data base is the same as for all previous tables and charts.
The premium to capital values are generally consistent with the underwriting
variability patterns noted previously. The general decline in the maximum ratio
over the 1972-1975 interval reflect the contrast between the profitable and unprofitable years of 1972 and 1975. While it has always been conceptually clear
that low combined ratios coupled with stable underwriting can permit premiums
to-capital ratios in excess of those under contrary positions, the extent and
magnitude of such differences had not always been clear. Table 12 helps to
estimate such differences,s2
The underwriting experience underlying the small premium volume classifications typically develops maximum ratios well below the 3.0 NAIC Early Warning Value and also the more conservative 2.0 Kenney rute.s3 Private Passenger
Auto Liability, Surety, and Liability Other Than Auto developed the lowes
ratios at small volumes.
Thus, for insurers just entering a new line of insurance, there would appear
to be few if any lines which could accommodate rapid premium growth withou
commensurate increases in surplus. For example, if the statutory capital requirement for entering private passenger auto liability were $500,000, premium
volumes in excess of that amount might raise ruin levels beyond those which
regulators were willing to tolerate. A firm which, for example, expected to write
$500,000 in private auto liability premiums might reasonably be expected t
maintain a capital of at least $500,000 or perhaps considerably more to absorb
underwriting fluctuations (assuming that a ruin assumption of 0.001 is reason
able). One of the interesting judgments to be made from the data of Table 12 is
the apparent continued relevance of the so-called Kenney rule to the monoline
underwriting activities of small insurers. The traditional "two to one" rule may
even be optimistic when applied to small premium volumes on a monoline basi
a situation more likely to exist with new insurers than with established ones.
Even more dramatic limitations on premium volume or increased capital migh
be expected of insurers entering the nonautomobile liability field, surety
fidelity and other lines capable of tolerating only modest ratios of premium to
capital.
S2For purposes of comparison, premiums-to-capital ratios were calculated for several
size groupings. Consistently similar size effects were observed throughout. Only with credit
insurance was higher volume associated with a lower ratio; underwriting variability at highe
volumes was relatively high. The line is very cyclical and highly subject to business conditions.
s3 Conversely, the values resulting from the experience at large volumes developed ratios
consistently in excess of the 3.0 value. In 1972, a year of record underwriting profit, larger
volumes frequently developed values in excess of 2.0 and 3.0 norms.
PROPERTY UNDERWRITING CAPITAL
HAMMOND-SHAPIRO
253
Insurers entering more than one line at a time might benefit from diversification and experience a more stable underwriting result. Although the crosssectional data of 1972 through 1975 cannot be used to forecast accurately the
underwriting experience of individual insurers, some insights into possible
diversification effects can be observed for the permissible premiums-to-capital
ratios associated with the line groupings of line 31 through line 39.~
The grouped data from Table 12 do not permit clear identification of
diversification effects because of the presence of size or pooling effects. Insurers
with more than one line of insurance, to remain in the smallest size category,
will have small premium volumes in each line thereby injecting a possible increase in variability to go along with the possible decrease associated with
diversification. The automobile lines (13, 14, 15 and 16) are grouped to form
line 35 which produces a higher premiums-to-capital ratio than for the separate
auto liability lines, but lower than for the physical damage lines - a kind of
leveling effect. Similar effects appear for other combinations of lines. It. is not
possible, however, to measure the extent in which such effects are produced
by diversification.
Summary and Conclusions
The principal problem in setting capital requirements for entrants is that
new insurers have no operating history. Underwriting results, management
philosophy, and the impact of economic conditions on capital must be assessed
on the basis of collective and not individual underwriting experience. Collective
results, however, are usually known only on an aggregate basis, thereby obscuring patterns and variations of relevance to regulatory assessments about capital
needs.
The cross-sectional measure of risk employed in this analysis measures the
variation in underwriting results across essentially all insurers operating in each
line of insurance from 1972 through 1975. It is suggestive of the regulatory
uncertainty surrounding the specification of capital for new insurers. Lines
which develop highly variable underwriting results produce a greater degree of
regulatory uncertainty about operating results than those which do not. It is
not unreasonable to think that entry of an insurer into such lines requires more
capital than entry into lines or classes of business where uncertainty is less.
Possible Regulatory Applications
The data on underwriting variability should help reduce regulatory uncertainty surrounding capital requirement specifications for new insurers. The
most important insights into underwriting risk assessment center upon apparent
size-effect problems for new insurers and differences in risk among lines.5s
54Technically, the extent to which an underwriting portfolio or more than one line or
insurance produces a more stable underwriting experience depends upon the intercorrelations
of underwriting results among insurance lines. The information required to study such
relationships is best developed from time series on individual firms. See Bachman,
Capitalization.
5SThese statements by themselves are not startling. Regulatory statutes, guidelines, and
regulations, however, frequently do not recognize these points.
254
REGULATION OF FINANCIAL INSTITUTIONS
Size. Small premium volumes are clearly associated with high underwriting
risk. While the study data do not provide conclusive evidence that the smaller
premium volume of a given insurer will produce more variation than higher
volumes, they strongly suggest it. The evidence supports the notion that small
insurers cannot, from a given capital base, write as much insurance as their larger
competitors and still maintain the same level of safety.
The information developed thus provides a reasonable basis for regulatory
interest in insurer expectations and market realities about premium volume
growth in relation to capital.
Risk Difference among Lines. Existing statutory capital requirements for
entry frequently specify different amounts for different lines of insurance. The
information developed in the study supports such differential requirements and
helps to identify lines of highest risk and therefore likely to require the highest
amounts of capital.
For small premium volumes, the lines which exhibited the greatest crosssectional risk were workers compensation, private passenger automobile liability,
and surety. These lines exhibited relatively high variability in each of the four
years under review. Lines developing high variability values in three of the four
years were: fidelity, air, ocean marine, group accident and health, boiler and
machinery, commercial automobile liability, and liability other than auto (the
latter including medical malpractice in all of the study years except 1975 as well
as product liability).
Changes in Statutes
The study interval is particularly interesting since it included a year of
record low combined ratios (1972) as well as a year of record highs (1975).
The data generally support regulatory and legislative judgment which implied
some lines as risky (e.g., surety, liability), thus requiring the greatest amount of
capital for entry.
The smaller premium volumes generally linked with new insurers appear
sufficiently associated with high underwriting risk to warrant conservative ratios
of premiums to surplus. The Kenney "two-to-one" rule appears optimistic when
applied to small premium volumes on a monoline basis.
Nearly all statutes specify a fixed amount of capital for entry into a line or
class of insurance. Yet, solvency does not depend on the absolute amount of
capital but on the premium volume in relation to capital. The expected premium
volume of new insurers, however, is affected by management philosophy, competition, profit, and market growth potential, conditions which may change
from one year to the next. These in turn may vary from one line to the next. A
statute which imposes only a fixed amount of capital cannot accommodate the
different risks associated with different ratios of premiums to capital.
The Wisconsin statute, however, by allowing regulatory discretion in the
establishment of entry capital amounts and by requiring premium growth plans
from prospective insurers (as part of the business plan), can deal with premiums
and capital together. The administration of such a statute is in turn facilitated
PROPERTY UNDERWRITING CAPITAL
HAMMOND- SHA PIR 0
255
by studies of underwriting risk and their continued development over time. It is
hoped that the analysis and information generated here will encourage the
adoption of statutes modeled after the Wisconsin Code and the continued
updating and possible refinement of the data.
Discussion
Robert C. Merton*
The two papers to be discussed, "Risk and Capital Adequacy in Banks" by
Sherman Maisel and "Capital Requirements for Entry into Property and Liability Underwriting: An Empirical Examination" by J.D. Hammond and Arnold
Shapiro, have much in common. Each paper is a summary of a larger study. Both
papers examine concern over the effects of regulation on competition among
intermediaries. The regulatory aspect that both focus onis the use of quantitative methods in the establishment of appropriate capital requirements. In measuring risk for the purposes of establishing capital requirements, both papers subscribe to the "portfolio approach" although the specific methodology employed
in each paper is somewhat different. Thus, I begin with some general points
which apply to the topics and approaches of both papers and leave for the end
specific points about each paper.
To put the analysis in these papers in perspective, it is useful to review briefly
why adequate capital is an important objective for the regulation of financial
intermediaries. Even the most elementary and abstract analysis of an economic
system would lead one to expect a widespread demand among individuals for
financial instruments which are functionally equivalent to bank deposits and
insurance. If they did not exist, then such contracts would have to be invented,
and hence, it is not necessary to dwell on why they exist. However, in a somewhat less elementary analysis, it can also be shown that the efficiency with
which these contracts perform their function is inversely related to their default
risk. The loss in efficiency from default risk is caused by significant increases in
both information and transactions costs. For example, the holder of a bank deposit which is known to be free of default risk requires little, if any, other information to understand the properties of the financial instrument he holds. However, once there is a possibility of default, then at a minimum, the holder of the
deposit must assess the probability of default and in the event of default, the
range of possible amounts that he might recover. Of course, to make such an
assessment requires data about the type and size of other liabilities of the bank;
the assets held by the bank and their risk; the operating expenses of the bank;
and the quality of its management. In addition, these data would have to be
analyzed to estimate the relevant probability distribution. Moreover, because
*Robert C. Merton is Professor of Finance at the Sloan School of Management at the
Massachusetts Institute of Technology. The author gratefully acknowledges aid from the
National Science Foundation.
256
DISCUSSION
MER TON
257
conditions within the bank change over time, such analyses would be required
frequently. In effect, the depositor must become a security analyst.
Even if independent firms evolved to provide such analytical services, the
depositor would still be left with the problem of evaluating the evaluators. And
in any event, prudence would dictate that he diversify by spreading his deposits
among many banks. Given the profile of a typical individual who uses bank deposits and the typical amounts involved per person, the aggregated information
and transactions costs caused by significant default risk are substantial. Along
the same lines, there is a serious loss in efficiency from default risk on insurance
contracts. While to some extent there is also a loss in efficiency from default
risk on general goods and services provided to individuals, the case of financial
intermediaries is special because to receive the services of an intermediary one
must become a liability-holder of the firm for the duration of the services and
because the aggregate amount of such "customer" liabilities is a significant
fraction of the total value of the firm. Therefore, "customer" liabilities of
intermediaries differ from the general liabilities of firms in that resources should
be spent to ensure that these customer liabilities are virtually default-free.
To ensure that these obligations to customers are met will in general require
a third party to set rules for performance by the intermediaries and to provide
surveillance to ensure compliance. It may also require that the third party
guarantee the customer liabilities. While it is not essential that this party be a
government agency, it is important that the capability and willingness of the
third party to meet its obligations be virtually beyond question. Otherwise, a
fourth party might be required to ensure the performance of the third one; and
a fifth for the fourth; and so on. The resulting "layering" of surveillance and
other costs are likely to be inefficient. Further, in the absence of detailed information and analysis, customers (e.g., depositors) may well use the size of the
guarantor as a selector for greater safety. This tendency could make it difficult
to maintain a competitive structure within which the private sector could provide these services at minimum cost. Of course, lack of competition will still be
a problem if a government agency provides this "third party" service. That is, if
a government agency provides the service, then the usual market forces which
tend to enforce efficient operations on the part of private enterprises will be
missihg. If the government agency also insures the liabilities (as with deposit insurance), then the absence of a market makes the determination of economically
sound insurance premiums especially difficult. Whether or not these services
should be provided by the private sector or government agencies is far from a
resolved issue in theory. However, as a practical matter, government agencies are
deeply involved in these activities, and at least, for large scale intermediation
such as in the banking system, the view that they should be is not without foundation.
If a government agency such as the Federal Deposit Insurance Corporation
does perform this function, then the information and diversification costs of
individual depositors are eliminated. Of course, this does not eliminate all such
costs since the agency must incur the costs of monitoring the performance of
the intermediaries. While common sense suggests that regulations be set so as to
keep these costs to a minimum, there are constraints on the form that these
258
REGULATION OF FINANCIAL INSTITUTIONS
regulations can take. For example, the cost to FDIC ensuring that deposit obligations are met could be minimized by simply requiring that all member banks
hold all their assets in short4erm, marketable U.S. government securities. While
this solution has at times been suggested, it would clearly eliminate the other important functional role for banks which is to make loans to businesses and individuals. The social cost of eliminating this service would most likely exceed the
benefit of reduced costs to FDIC. Thus, the regulations should be chosen so as
to minimize the costs of insolvency subject to the constraint that the regulations
do not significantly impair the intermediaries’ capabilities to perform their functional roles.
While "insolvency" has been defined in a variety of ways, the definition
used by Maisel (with some slight modification) is an appropriate one for the
purpose of regulation. In the Maisel paper, insolvency of an intermediary is said
to occur "when the market value of its liabilities exceeds that of its assets reduced by the costs of bankruptcy." It should be noted that "liabilities" as used
by Maisel refer to " ’customer’ liabilities" (as I have described them) and not the
general liabilities of the intermediary which would include equity. While I
applaud his emphasis on "market" rather than "book" values for determining insolvency, his definition has a technical difficulty because the limited-liability
feature of equity implies that the market value of equity (and hence, net worth)
cannot be negative. However, if his definition is modified to read "when the
value of its liabilities (to consumers), computed by assuming that the terms of
such obligations wouM be fully met, exceeds the market value of its assets reduced by the costs of bankruptcy," then net worth, defined as the difference
between the value of assets and the value of liabilities computed in this way, cart
be negative. Under this definition, negative net worth implies insolvency, and its
probability is represented by the shaded area of the probability distribution for
net worth as illustrated in Figure 1 of the Maisel paper.
The risk of insolvency will depend upon the volatilities of the intermediary’s
assets, liabilities, and operating costs. It will also depend upon the frequency
with which the intermediary is evaluated by the regulators, and on the amount
of capital or assurance money provided by the equityholders of the intermediary.
Regulatory restrictions can be imposed on all of these items to reduce the risk of
insolvency. However, as discussed earlier, there are limits to the restrictions
which can be placed on the volatility of either assets or liabilities without significantly interfering with the functions served by the intermediary. It is for this
reason that both papers focus on the establishment of adequate capital requirements. In the case of the insurance industry examined by Hammond and Shapiro,
the capital adequacy requirement is indeed the central device for protecting
policyholders. For the banking industry where most deposits are insured, the
central protection is of course deposit insurance. However, the capital requirement is the central device for protecting the insurer of those deposits (e.g., FDIC
or FSLIC). The insurance premium charged by FDIC is also an important control
at least in principle. That is, banks with smaller amounts of capital or riskier
assets could be charged a higher premium. However, as Maisel points out, FDIC
charges all banks the same insurance rate independent of their risk, and therefore, under current practices, this second control is not very effective.
DISCUSSION
MER TON
259
The appropriate capital requirement to meet a specified level of insolvency
risk will depend upon the volatility of both the intermediary’s assets and its
liabilities. Both papers fall short of a complete analysis in this respect because
Maisel only studies the volatility of the assets and Hammond and Shapiro only
study the volatility of the liabilities and operating costs. However, both papers
do stress that it is the riskiness of the portfolio (of assets or liabilities) which is
important for establishing capital requirements. While this approach will come
as no surprise to students of modern portfolio theory, it is in sharp contrast to
the traditional treatment of risk in regulation where the practice is to set risk
limits on each individual component of the (asset or liability) portfolio held by
the intermediary. As the authors of both papers point out, this traditional
approach explicitly neglects the important role diversification plays in the reduction of risk. The analyses presented in the Hammond and Shapiro paper provide
empirical evidence of the important benefits from diversification across multiple
lines of insurance. Further evidence of the importance of diversification can be
found in parallel studies on the regulation of money-management fiduciaries and
revisions of the "Prudent Man" rules. And indeed, recent guidelines for ERISA.
type pension accounts suggest a move away from the traditional view of setting
risk standards for each investment in the account and a move toward replacing
these standards with ones applied to the account as a whole.
In both papers, the authors chose to measure risk by variance (or equivalently, standard deviation). This choice raises two questions: Should one use
"total" risk or only its "systematic" component as the appropriate measure? If
"total" risk is the correct measure, does the variance adequately measure it? In
his paper, Maisel discusses both questions and I agree with iris answers. On the
first, because both papers are concerned with measuring default risk, total risk
is the proper measure as has been shown elsewhere.~ On the second, the answer
is less clear. If the dynamics of asset and liability value changes are such that
they can reasonably be modeled by diffusion processes, then the variance rate
will be an adequate statistic. However, if these dynamics involve radical changes
in value over a short period of time, then variance will not be sufficient, and
more complex measures such as those associated with either stable or Poisson.
directed distributions may be required. While the empirical resolution of this
question would be important prior to the implementation of either paper’s
quantitative methods as a formal part of the regulatory process, the qualitative
indications for regulatory change as suggested by either paper would be largely
unaffected.
I now turn to the specifics of each paper beginning with Maisel’s.
Research in finance theory has produced a number of important quantitative tools for analyzing risk and evaluating insurance premiums. These quantitative methods have been subjected to empirical verification and many have be1 See, for, example, R.C. Merton, 1974, "On the Pricing of Corporate Debt: The Risk
Structure of Interest Rates," Journal of Finance 29, pp. 449-70 and 1977, "An Analytic
Derivation of the Cost vf Deposit Insurance and Loan Guarantees," Journal of Banking
and Finance 1, pp. 3-11.
260
REGULATION OF FINANCIAL INSTITUTIONS
come standard operating tools in the (private sector) financial community. I am
strongly in agreement with Maisel’s central theme that these tools should be employed within the bank regulation and examination system to provide more
objective evaluations. Indeed, because government agencies such as FDIC and
FSLIC do not have a competitive market providing price data on their type of
insurance premiums and because they do not have publicly traded share prices to
indicate how the market perceives these agencies’ performance, the use of these
quantitative tools for making objective evaluations may be more important for
these agencies than they would be for a corresponding private-sector firm.
Although the issues raised and analyses presented by Maisel are confined to
the banking system and deposit insurance, his contribution has added significance because much of what he has done can be applied with minor modification to other areas of public-sector guarantees. Two such areas of particular importance are government loan guarantees and government guarantees of private
pension obligations. Both areas promise to be topics for much future discussion,
and the role of quantitative methods in making objective evaluations of the costs
and benefits will be at least as important in these areas as it is for the banking
system.
The Maisel paper covers most of the important issues. He recognizes that the
establishment of "adequate" capital requirements cannot be undertaken without
a simultaneous analysis of bank asset risks and the schedule of insurance premiums charged by FDIC. In his section titled "Measuring Capital Adequacy in a
Bank," he correctly stresses the importance of using "economic, rather than
book or reported capital" in assessing the risk of insolvency. He underscores
the empirical significance of this point in the section titled "Measuring Net
Worth" where he describes the historically large discrepancies between the
market and book values of net worth for large banks. I agree with the conditions
he describes for "capital to be adequate" in that section although I would
replace his description of a "fair" insurance premium as one that "... covers the
expected losses of the insurer ..." with "... compensates fully the insurer in
terms of ex-ante expected return for the risks borne...". The reason for this suggested change is that some of the risk borne by the insurer may be market-related
or systematic risk in which case the insurer should be compensated by an expected return in excess of the risk-free rate, and therefore, the premium should
cover more than expected losses. It should also be mentioned that the probability of insolvency is not a sufficient measure of the risk of insolvency because
it does not capture the magnitudes of the losses in the event that insolvency
Occurs.
In his analysis of bank asset risk (in "Finding the Risk in a Bank"), Maisel
lays out the various categories of risk and their relative importance. By emphasizing broad risk classes, he incorporates the effects of diversification and correctly points out that the important risks are the nondiversifiable ones such as
interest rate risk. He also points out the importance of taking into account
"off-balance-sheet" liabilities such as credit lines and "stand-by" agreements to
purchase mortgages in evaluating the bank’s risk position. Since these liabilities
are essentially option-type financial contracts, the powerful tools developed in
finance theory for pricing options could be especially useful in their evaluation.
DISCUSS ION
MER TON
261
In discussing moral hazard risks, Professor Maisel states that "Among banks
as a whole, the greatest risks and most common cause of failure are due to fraud,
either internal or external, and to insider abuse." As one might expect, these
moral hazard risks are more important among smaller banks, and therefore, they
may not represent a serious threat to the banking system as a whole. Moreover,
while it would be naive not to provide safeguards against such abuses, these
abuses often violate the criminal code, and therefore, vigorous enforcement of
this code may possibly provide the best protection. However, there is another,
far more important, moral hazard problem which Professor Maisel discusses in
the section titled "Fair Insurance Premia." Namely, by insuring the deposits,
FDIC "induces" the banks to pursue more risky investment strategies. Maisel
describes the problem in terms of the current system as "A flaw in the present
system lies in the fact that banks may find it profitable to increase their risks,
since there is only a slight relationship between risks and their costs of insurance.
This can lead to a constant losing battle by regulators to force specific banks to
reduce their risks."
To solve this problem, Maisel indicates that a system of variable rates would
be feasible. While variable rates based upon differences in risk among banks is
preferable to a uniform rate, variable rates alone will not solve the problem. If
the rate is set on the basis of an ex ante assessment of the bank’s risk, then there
is still an incentive for the bank to "cheat" by following ex post, a more risky
investment strategy. To deal with this issue, Maisel suggests that "Adjustments
in ratings and charges could be made retrospectively to guard against major shifts
in operations." The effectiveness of this method is of course an empirical issue,
and it will certainly depend upon the frequency and care with which FDIC
monitors the risk position of each bank. However, my belief is that such adjustments will not be adequate. Given the level of premiums currently charged
by FDIC, even with some adjustment, the "cost" to a bank from pursuing
a riskier investment strategy appears small. Indeed, if a bank "wins" on these
riskier investments, then it would probably be more than happy to pay the
additional assessment. If it "loses," then how does FDIC collect? While a definitive solution to this problem has not as yet been presented, one possible avenue
for exploration would be to replace (or at least supplement) the current practice
of annual charges with a large "front-end" er~try fee for membership in FDIC. As
I have shown elsewhere,~ this type of charge will reduce (and under some conditions, eliminate) the "FDIC-induced" incentives for a bank to pursue a riskier investment strategy. Of course, like larger initial capital requirements, such charges
would make entry into the banking industry more costly, and therefore, may
have a negative effect on competition.
In summary, the Maisel paper does not resolve all the theoretical problems
involved in bank regulation and deposit insurance. There certainly remain many
practical difficulties in developing the "standardized" methods of evaluation and
the necessary supporting data to implement the procedures recommended. How~ See R.C. Merton, 1978, "On the Cost of Deposit Insurance When There Are Surveillance Costs," Journal of Business 51, pp. 439-452 and especially pages 447-450.
262
REGULATION OF FINANCIAL INSTITUTIONS
ever, these problems and difficulties can be solved, and I firmly believe that the
lines suggested by Maisel are in the right direction toward a vastly improved
system.
The Hammond and Shapiro paper on insurance regulation is more narrow in
focus than the Maisel paper. Specifically, they concentrate on entry capital
requirements for nonlife insurance companies, and their principal contribution is
to provide empirical estimates of differences in risks among various insurance
lines and to demonstrate the benefits of diversification in multiple-line activities.
Their quantitative analyses permit an evaluation of the current "rule-of-thumb"
practices in setting entry capital and suggest directions for change in entry capital requirement statutes. In their analysis of the appropriate entry capital requirement, they consider only the underwriting and operating expense risks and
not the risks associated with the assets held by the insurance company. For
many of the same reasons given in footnote five of their paper, it would appear
that the risks of the assets held by fledgling insurance companies should be an
integral part of the determination of entry capital. In particular, the relatively
small size of new insurance companies’ asset bases may make them especially
vulnerable to significant risks from insufficient asset diversification.
On the whole, the empirical analysis presented is comprehensive and staffstically proper. The use of a combined time series and cross-section analysis of
the data is especially to be applauded. However, I have some concern with the
use of the combined ratio for the purposes here even though it is the standard
practice to use it as the measure of underwriting performance. For example, the
authors find that smaller (and therefore, presumably newer) insurance companies have a much larger standard deviation of underwriting performance than
do larger (and therefore, presumably more established) companies. It seems to
this reader that it would be useful for regulatory purposes to know whether this
higher variation was principally due to the expense ratio or the loss ratio. If it
were the expense ratio, then do these results suggest greater variation in management skills available to the smaller companies? Or do they suggest that new companies tend to enter into the "tougher" part of the market where expenses are
more uncertain? Or are these differences principally the result of accounting
"biases" which differentially affect smaller, newer, or faster growing insurance
companies but which tell us very little about the relative risks of default between
small and large companies? If it were the loss ratio, then do these results suggest
that the premiums charged by smaller firms are more variable? Or is it that
smaller companies attract riskier customers even within the same insurance line?
By separating the two ratios, the authors can verify that a principal source
of greater risk to smaller companies is the lack of diversification among customers
within a product line. It is well known that for independent customers of about
the same size, the variation in the loss experience should be roughly proportional
to one-over-the square root of the number of customers. There is no strong
reason to expect that expense ratios should similarly benefit from such customer
diversification, and indeed, the expense ratio might even increase.
These comments and questions should not be interpreted as a negative
report on the paper. Many of these questions may be answered in their cited
DISCUSSION
MER TON
263
larger study or in the works of others. It is also clear that some of the data required to answer these questions may not be available. Moreover, the use of the
combined ratio causes fewer problems for the purpose of distinguishing risk
differences among insurance lines, and the analysis presented should be helpful
to insurance regulators. However, on the specific regulatory issue of entry
capital requirements, an expanded analysis of the risk characteristics of smaller
insurance companies would seem to be in order.